Long-Term Optimization Logic: Traditional

Long-Term Optimization studies are used to forecast capacity expansion resources and retirements. Future resources can be included in the Resources table with pre-determined start dates. Or, the New Resources table can be used in conjunction with the long-term optimization logic to use market economics to select resource additions and retirements. This optimization process simulates what happens in a competitive marketplace and produces a set of future resources that have the most value in the marketplace.

Aurora assumes that new generators will be built (and existing generators retired) based on economics. In the traditional method, the economic measure used is real levelized net present value (revenues less cost) on a $/MW basis. Investment cost is included in the cost portion of the formula. Also, the methodology assumes that potentially non-economic contracts will not influence the marketplace and that someone will capture the opportunity value of non-economic contracts. Therefore contracts are not modeled in the pricing component of Aurora. (See also Mixed-Integer Program Logic for LT article for alternate methods to the traditional logic.)

In preparing for a Long-Term Optimization study, identify New Resources to be evaluated and determine parameters for the study.

New Resources

The New Resources Table in the input database is used to define a new resource and its operating characteristics. For example, the type of resource (e.g., wind, solar, nuclear, coal, gas, etc.). The New Resources Table contains columns used to define all the assumptions for a new unit, including when the potential unit will be placed in service. These assumptions provide controls for placing operating constraints on all the units in the system.

Aurora will calculate a value for each potential new unit. This value is the real levelized net present value (NPV) in $/MW, and includes capital costs. Aurora uses the real levelized cost to make decisions about new units and resource retirements.

Resource Value

Aurora determines resource value from the difference between market price and resource cost. This determination is performed for every hour for every resource in the region. Thus, a very accurate value is developed which takes into account system value during all time periods (i.e., on-peak, off-peak and other hours; and during daily, seasonal, and annual periods).

The user can specify the use of variable operation and maintenance expenses along with fixed operation and maintenance expense in the computation. We recommend however, that the value reporting be performed on all forward costs. This produces the best economic view of the resource.

Steps in Traditional Long-Term Optimization

See the Aurora Long Term Optimization Flow Diagram for visual representation. Also see Mixed-Integer Program Logic for LT article for alternate methods to the traditional logic.

Aurora uses the following steps in the Traditional Long-Term Optimization (Capacity Expansion) process:

The first iteration begins with resources selected to meet the planning reserve margins for the zones and pools being run. If reserve margin targets are not being used, the model will assume a reserve margin of the minimum of 0% as the beginning first year reserve margin for each pool and zone. The model will make the first iteration build decisions based on the new resource fixed costs.
Aurora enumerates all new resources.
The value for each existing resource is determined.
The value for each new enumerated resource is determined.
Resources are sorted by value.
A small set of the resources with the most negative values is selected to retire (note that no retirements will be selected after the first iteration)..
A small set of new resources with the most positive values is selected to add.
Aurora is re-run to compute new electric prices and resource values.
The process is repeated until the system stabilizes.

This is done on a gradual basis using small sets of changes because large changes to resources would change all of the assumptions used to compute value. This optimization approach provides an excellent approximation for how the competitive marketplace will select resources in the long-term. Resources that create value on a going-forward basis will be constructed while those that have no value on a going-forward basis will be retired.

The primary result of the Long-Term Optimization analysis is creation of the Resource Modifier Table, which becomes part of the Aurora input database. It will then be used to add or retire resources in the Resources Table. This table is the only output saved to the input database.

Simulation Logic

Long-Term Optimization Logic